You can have the LLM itself generate it based on the documentation, just like a human early adopter would